This month marks a decade since Netflix - the world's most influential and widely subscribed streaming service - launched in Australia.
Since then the media landscape has undergone significant transformation, particularly in terms of how we consume content. According to a 2024 Deloitte report, Australians aged 16-38 spent twice as much time watching subscription streaming services as free-to-air TV (both live and on-demand).
Part of the success of streaming services lies in their ability to provide content that feels handpicked. And this is made possible through the use of sophisticated recommender systems fuelled by vast amounts of user data.
As streaming viewership continues to rise, so too do the risks associated with how these platforms collect and handle user data.
Changing methods of data collection
Subscription streaming platforms aren't the first to collect user data. They just do it differently.
Broadcasters have always been invested in collecting viewers' information (via TV ratings) to inform promotional schedules and attract potential advertisers. These data are publicly available.
In Australia, TV data are collected anonymously via the OzTam TV ratings system, based on the viewing habits of more than 12,000 individuals.
Each television in a recruited household is connected to a metering box. Members of the household select a letter that corresponds to them, after which the box records their viewing data, including the program, channel and viewing time. But this system doesn't include broadcasters' video-on-demand services, which have been around since the late 2000s (with ABC iView being the first).
In 2016 a new system was launched to measure broadcast video-on-demand data separately from OzTam ratings.
However, it collected data in a rolling seven-day report, in the form of total minutes a particular program had been watched online (rather than the number of individuals watching, as was the measurement for TV). This meant the two data sources couldn't be combined.
In 2018, OzTAM and Nielsen announced the Virtual Australia (VoZ) database which would integrate both broadcast TV and video-on-demand data. It took six years following the announcement for the VoZ system to become the industry's official trading currency.
Streamers' approach
Streaming platforms such as Netflix have a markedly different approach to acquiring data, as they can source it directly from users. These data are therefore much more granular, larger in volume, and far less publicly accessible due to commercial confidence.
In recent years, Netflix has shared some of its viewing data through a half-yearly report titled What We Watched. It offers macro-level details such as total hours watched that year, as well as information about specific content, including how many times a particular show was viewed.
Netflix also supplies information to its shareholders, although much of this focuses on subscriber numbers rather than specific user details.
The best publicly accessible Netflix data we have is presented on its Tudum website, which includes global Top 10 lists that can be filtered by country.
The main data Netflix doesn't share are related to viewer demographics: who is watching what programs.
Why does it matter?
Ratings and user data offer valuable insights to both broadcasters and streaming services, and can influence decisions regarding what content is produced.
User data would presumably have been a significant factor in Netflix's decision to move into live content such as stand-up comedy, the US National Football League (NFL) and an exclusive US$5 billion deal with World Wrestling Entertainment.
Streaming companies also use personal data to provide users with targeted viewing suggestions, with an aim to reduce the time users spend browsing catalogues.
Netflix has an entire research department dedicated to enhancing user experience. According to Justin Basilico, Netflix's Director of Machine Learning and Recommender Systems, more than 80% of what Netflix users watch is driven by its recommender system.
As noted in its privacy statement, Netflix draws on a range of information to provide recommendations, including:
- the user's interactions with the service, such as their viewing history and title ratings
- other users with similar tastes and preferences
- information about the titles, such as genre, categories, actors and release year
- the time of day the user is watching
- the language/s the user prefers
- the device/s they are watching on
- how long they watch a particular Netflix title.
If a user isn't happy with their recommendations, they can try to change them by editing their viewing and ratings history.
Personalised or predetermined?
The rise of streaming hasn't only transformed how we watch TV, but also how our viewing habits are tracked and how this information informs future decisions.
While traditional broadcasters have long relied on sample anonymised data to measure engagement, streaming platforms operate in a landscape in which detailed user data can be used to shape content, recommendations and business decisions.
While personalisation makes streaming more appealing, it also raises important questions about privacy, transparency and control. How much do streaming platforms really know about us? And are they catering to our preferences - or shaping them?
Marc C-Scott, Associate Professor of Screen Media | Deputy Associate Dean of Learning & Teaching, Victoria University
This article is republished from The Conversation under a Creative Commons license. Read the original article.
https://counter.theconversation.com/content/244921/count.gif?distributo…" alt="The Conversation" width="1" height="1" referrerpolicy="no-referrer-when-downgrade" />
